Abstract: The amino acid sequences of caffeine synthase from Camellia sinensis, Theobroma cacao, Camellia japonica and other plants which were registered in GenBank, were analyzed and predicted by bioinformatic tools in subsequent aspects, including the isoelectric point, subcellular localization, signal peptide, transmembrane topologieal structure, conserved functional domain, motif, secondary structure and tertiary structure of protein.Resultsshowed that the caffeine synthase of plants which were located in cytoplasm and nuclei, and had phosphorylation, acylation, glycosylation sites could be divided into three different types based on gene sequences and conservative domains.Two of them, type I and type II protein, wereα-type soluble proteinases, and the secondary structure of type III proteinase was rich in random coil and has potential signal peptide, but they all did not have transmembrane helical structure.The result of tertiary structure prediction indicated that type I protein and type II protein were similar, they were all composed ofα-helix and horizontalβ-folded layers, but in the type III protein theα-helixes locateed in the lateral ends and were connected by verticalβ-folded layers.